refactor(api): update AutoRAG search parameter handling to only include defined values in RagAiSearchNode and RagSearchNode

Author: bchapuis

apps/api/src/nodes/rag/rag-ai-search-node.ts

@@ -2,6 +2,7 @@ import { NodeExecution, NodeType } from "@dafthunk/types";
 
 import { ExecutableNode } from "../types";
 import { NodeContext } from "../types";
+import { createDatabase, getDataset } from "../../db";
 
 /**
  * RAG AI Search node implementation
@@ -110,8 +111,15 @@ export class RagAiSearchNode extends ExecutableNode {
         return this.createErrorResult("AI binding is not available");
       }
 
+      // Verify dataset exists and belongs to organization
+      const db = createDatabase(context.env.DB);
+      const dataset = await getDataset(db, datasetId, organizationId);
+      if (!dataset) {
+        return this.createErrorResult("Dataset not found or access denied");
+      }
+
       // Create multi-tenant folder filter
-      const folderPrefix = `${organizationId}/${datasetId}/`;
+      const folderPrefix = `${datasetId}/`;
 
       // Prepare AutoRAG search parameters - only include defined values
       const searchParams: any = {
@@ -150,7 +158,7 @@ export class RagAiSearchNode extends ExecutableNode {
         searchParams.filters = {
             type: "eq" as const,
             key: "folder",
-            value: "01975f74-76b4-7778-b50d-1d079cee26ce/01975fd5-5e25-728a-afd3-a8c40e6571b7/",
+            value: folderPrefix,
         };
       }
 

apps/api/src/nodes/rag/rag-search-node.ts

@@ -2,6 +2,7 @@ import { NodeExecution, NodeType } from "@dafthunk/types";
 
 import { ExecutableNode } from "../types";
 import { NodeContext } from "../types";
+import { createDatabase, getDataset } from "../../db";
 
 /**
  * RAG Search node implementation
@@ -98,8 +99,15 @@ export class RagSearchNode extends ExecutableNode {
         return this.createErrorResult("AI binding is not available");
       }
 
+      // Verify dataset exists and belongs to organization
+      const db = createDatabase(context.env.DB);
+      const dataset = await getDataset(db, datasetId, organizationId);
+      if (!dataset) {
+        return this.createErrorResult("Dataset not found or access denied");
+      }
+
       // Create multi-tenant folder filter
-      const folderPrefix = `${organizationId}/${datasetId}/`;
+      const folderPrefix = `${datasetId}/`;
 
       // Prepare AutoRAG search parameters - only include defined values
       const searchParams: any = {

apps/api/src/nodes/types.ts

@@ -105,6 +105,7 @@ export interface NodeContext {
   httpRequest?: HttpRequest;
   emailMessage?: EmailMessage;
   env: {
+    DB: D1Database;
     AI: Ai;
     AI_OPTIONS: AiOptions;
     RESSOURCES: R2Bucket;

apps/api/src/routes/datasets.ts

@@ -203,7 +203,7 @@ datasetRoutes.post("/:id/upload", jwtMiddleware, async (c) => {
     }
 
     // Create the R2 path following the multitenant pattern
-    const r2Path = `${organizationId}/${id}/${file.name}`;
+    const r2Path = `${id}/${file.name}`;
 
     // Upload to R2
     await c.env.DATASETS.put(r2Path, await file.arrayBuffer(), {
@@ -243,7 +243,7 @@ datasetRoutes.get("/:id/files", jwtMiddleware, async (c) => {
 
   try {
     // List objects in the dataset's directory
-    const prefix = `${organizationId}/${id}/`;
+    const prefix = `${id}/`;
     const files = await c.env.DATASETS.list({ prefix });
 
     const response: ListDatasetFilesResponse = {
@@ -277,7 +277,7 @@ datasetRoutes.delete("/:id/files/:filename", jwtMiddleware, async (c) => {
   }
 
   try {
-    const r2Path = `${organizationId}/${id}/${filename}`;
+    const r2Path = `${id}/${filename}`;
     await c.env.DATASETS.delete(r2Path);
 
     const response: DeleteDatasetFileResponse = {
@@ -308,7 +308,7 @@ datasetRoutes.get("/:id/files/:filename", jwtMiddleware, async (c) => {
   }
 
   try {
-    const r2Path = `${organizationId}/${id}/${filename}`;
+    const r2Path = `${id}/${filename}`;
     const object = await c.env.DATASETS.get(r2Path);
 
     if (!object) {

apps/api/src/runtime/runtime.ts

@@ -318,6 +318,7 @@ export class Runtime extends WorkflowEntrypoint<Bindings, RuntimeParams> {
         emailMessage,
         onProgress: () => {},
         env: {
+          DB: this.env.DB,
           AI: this.env.AI,
           AI_OPTIONS: aiOptions,
           RESSOURCES: this.env.RESSOURCES,